Statistics Netherlands

Directorate of Socio-economic and Spatial Statistics

P.O. Box 4481, 6401 CZ Heerlen

The Adult Education Survey (AES) covers adults’ participation in education and training (formal - FED, non-formal - NFE and informal learning - INF). The 2022 AES focuses on people aged 18-69. The reference period for the participation in education and training is the twelve months prior to the interview.

Information available from the AES is grouped around the following topics:

• Participation in formal education, non-formal education and training and informal learning
• Volume of instruction hours
• Characteristics of the learning activities
• Reasons for participating
• Obstacles to participation
• Access to information on learning possibilities and guidance
• Employer financing and costs of learning
• Self-reported language skills

For further information see the 2022 AES legislation (http://ec.europa.eu/eurostat/web/education-and-training/legislation) and the 2022 AES implementation manual (http://ec.europa.eu/eurostat/web/education-and-training/methodology).

- Classification of Learning Activities (CLA, 2016 edition)
- International Standard Classification of Education 2011 (ISCED 2011)
- Classification of Occupations 2008 (ISCO 08)
- Classification of economic activities Rev. 2 (NACE Rev. 2)

AES covers all economic sectors.

Definitions as well as the list of variables covered are available in the 2022 AES implementation manual (http://ec.europa.eu/eurostat/web/education-and-training/methodology).

Individuals, non-formal learning activities.

Individuals aged 18-69 living in private households.

The Caribbean Netherlands are excluded. 

The current wave of the AES was carried out in 2023.

Fieldwork period: 10 January to 31 March, 2023.

Not applicable.

Number, EUR.

The fieldwork was carried out in the period January 10th to March 31st 2023. The reference period is January 2022 to March 2023.

At European level:

Basic legal act: Regulation (EU) 2019/1700

Implementing act: Commission Implementing Regulation (EU) 2021/861

At national level:

No legal acts or agreements

Not applicable.

CBS is bound by the European General Data Protection Regulation (GDPR). This Regulation helps protect the privacy of citizens. In addition, CBS adheres to the privacy stipulations in the Statistics Netherlands Act, the European Statistics Code of Practice, and its own Code of conduct (Dutch only).

Under certain conditions, Statistics Netherlands’ Centre for Policy Related Statistics (CvB) makes available microdata (anonymous data at the level of individual persons and businesses) for statistical research. Usage of these data requires the organisation the researcher works for to be authorised by the Advisory Board. The development of our microdata services is done in consultation with our customers board, a representative group of our users. The researcher can conduct his research on site at our offices in The Hague and Heerlen, or from your own workplace using a secure internet connection (remote access). In addition to existing available microdatasets, other datasets can be made available. Custom-made datasets with Statistics Netherlands microdata can be compiled and it is also possible to use your own datasets. The latter will be prepared by CvB so they can be linked to our datasets. All datasets you use remain on the dedicated server at Statistics Netherlands. Statistical results (e.g. statistical tables) can be relocated by CvB staff so you can use them outside our secure environment.

Personal data are anonymised by deleting variables that can reveal identification of persons and unique identification key is used for matching. 

Before statistical results are released, we check all data for the risk of disclosure, as Dutch law forbids Statistics Netherlands to disclose identifiable information on individual persons, enterprises, institutions or households. Statistics Netherlands requires that all your statistical results be published and made available to other interested persons and organisations. We publish an overview of publications based on research based on microdata from Statistics Netherlands. All information about our microdata services and the costs involved can be found in the Catalogue of services and tariff structure which is available on the Statistics Netherland's website (www.cbs.nl).

The CBS publication calendar lists the publication dates and times of all upcoming CBS news releases, when research results are made public. All statistics are also published as tables in the StatLine databank and released at or after midnight (00:00 hours). 

A press release and tables based on the new AES data are scheduled for the first quarter of 2024. Exact dates are not yet known.

http://www.cbs.nl/en-gb/

https://www.cbs.nl/en-gb/publication-calendar

Statistics Netherlands (CBS)’ mission is to compile official statistics and to publish the results. CBS makes these results accessible to the public through various channels. An outline of CBS’ publication policy can be found here. On this subject, the Advisory Council has issued an advisory report.

CBS pre-release access policy

Prior to publication of a news release, CBS reserves the right to grant pre-release access under embargo to relevant government departments, institutions and news organisations. The pre-release access policy is described in a memorandum on this website.

Every 6 years.

Ad-hoc news releases when required. 

There have not been any press releases on AES data yet. 

A press release is scheduled for the first quarter of 2024.

• Statline (each new edition of AES, once in six years)
• Press release (each new edition of AES, once in six years)
• Tables on request of local and national authorities (i.e. department of Employment and Social affairs; department of Education, Culture and Science)

One or more of the following publications may be carried out: 

• Statistische Trends (one or several articles)
• Themepages about education and the labour market
• Onderwijs in Cijfers (Education in figures) (once in six years)
• Chapter in ICT, Kennis en economie (ICT, Knowledge and economy) (once in six years)

Our data will become accessible in StatLine in the first quarter of 2024. 

Not applicable.

Our micro data are accessible under certain conditions: The criteria for research institutions who want to have access to the micro-data are:

Declination of these criteria is possible in certain cases. However, the Advisory Board needs to motivate this deviation explicitly. 

1. the institution has an independent legal individuality, or belongs to a public service.

On this last case you can think of a statistical or research department of ministries, provinces or municipalities.

These research institutions have to make a reasonable case for executing their research task under the condition mentioned under 2.

2. the institution isn't subjected to the authority of an administrative body.

3. the institution has to have a primary research aim.

4. the institution has to publish publicly. If the institution executes their research for a customer then the institution has to make public the results for which the data are used.

5. the institution enjoys a good reputation.

http://www.cbs.nl/en-gb/

https://www.cbs.nl/en-gb/publication-calendar

Not applicable.

Research description Adult Education Survey (AES) (in Dutch)

This description will be updated before national publication of the AES 2022 data.

Information about access to microdata AES2016 (in Dutch)

Tables in Statline (in Dutch):

Cursusdeelnemers bedrijfskenmerken, AES2016

Cursusdeelnemers persoonskenmerken, AES2016

For a detailed description of methods and concepts used, as well as for other documents related to the AES, please consult the reference documents related to 2007, 2011, 2016 and 2022 AES on CIRCABC. 

Not applicable.

See item 11 below. 

The following activities were implemented in NL: 

• the use of a web based questionnaire. This avoids routing issues and obliges respondents to answer in a consistent way. This improves the quality of the received data.
• emphasis on guaranteed privacy;
• clear questionnaire;
• possibility to contact a helpdesk;
• sending 2 reminders;
• use of registers where possible to collect the data (e.g. household income).

Strong points:

• AES in CAWI mode seems effective. As far as we can see at this stage, the quality of data is reasonable.
• For the overlapping concepts in LFS and AES (background characteristics and participation in formal education) we used the formulation of questions as it was in the LFS 2022. This turned out to be efficient in the development phase of the questionnaire. 
• Much attention in the questionnaire development was paid to produce a user-friendly survey: questions that are well understood for respondents.
• On a related note, the questionnaire was also developed with the goal-variables in mind. Specific concepts and variables that were very hard to deduct in past iterations of the AES were relatively easy to deduct in the current iteration since the questionnaire more closely resembles the prescribed concepts. It is always difficult to balance respondent-friendliness and researcher-friendliness in the development of a questionnaire, but we feel that we hit a good balance in this iteration of the AES.
• Necessary work was also done in the Question Lab test to adapt the questionnaire to modern needs. For the first time, the AES can now also be completed on a smartphone or tablet.
• The response percentage of CAWI was larger than expected.
• The data collection is carried out in accordance with the set time schedule.

Weaknesses:

• In general, like AES 2016, the survey duration was acceptable (around 20 minutes). However, specifically for respondents who participated in multiple educational activities the questionnaire was rather long.
• Long reference period can be problematic if asking detailed questions about a randomly selected NFE-activity that took place early in the reference period. As a consequence, e.g. the variables that contain estimates of costs and time spent related to learning activities are probably connected with measurement errors. 
• The concept of selection of random NFE activities based on respondent input for the purpose of follow-through questioning leads to a complex questionnaire. Complex processing is needed to arrive at Eurostat indicators based on the collected data. This is an intensive process.
• The process of specifying the variables requested by Eurostat: The variables specified by Eurostat almost always come with a filter. This filter describes the population of the variable. Any respondent who is not described in this filter is to be coded as “Not applicable”, with a value of “-2”. In some cases however, a situation occurs where the result of a respondent within the context of this variable can be best described as “Not applicable”, while still adhering to the filter. This can, for example, occur when we expand on an existing question because we feel that the prescribed question is not satisfying for respondents to answer. An example of this is the variable NFEONMAT. Because online lessons contain a lot of factors that are not always applicable to all respondents, we have expanded the questions for online lessons by including an answering category for “Not applicable”, to ensure that respondents will always have something to answer. But within the context of the variable NFEONMAT, answering “Not applicable” on the question cannot be specified as a “Not applicable” on NFEONMAT, since that option is reserved for the filter. Hence, we have coded respondents who give this answer as “-1” for “Not stated”, since we lack information to deduct any of the other answering-categories, or because we cannot specify which of the aspects of the question is “Not applicable” (Is the existence of online lessons not applicable, are the teachers not applicable or is the sharing of materials not applicable?).

Findings based on the questionnaire Lab test:

• For those respondents in formal education, the order of questions comes across unnaturally.
• For students and students in paid internships some job related questions were experienced as less appropriate. This younger age group (18-24 years) differs substantially from the group 25-69 years. 

Before the survey was carried out, we identified what the needs of national policy makers were with regard to the topics in the AES. See items 12.1 and 12.2.

Policy makers at national level (e.g. ministries) - topics of interest:

• Access to information on learning possibilities and guidance
• Participation in education and training by type, characteristics of the activity (field, distance learning, etc.)
• Reason, use and outcomes of FED and NFE
• Share of job-related or employer-sponsored NFE
• Volume of instruction hours for FED and NFE
• Cost of learning for NFE
• Obstacles to participation in education and training

Media - topics of interest:

• Access to information on learning possibilities and guidance
• Participation in education and training by type, characteristics of the activity (field, distance learning, etc.)
• Reason, use and outcomes of FED and NFE
• Share of job-related or employer-sponsored NFE
• Volume of instruction hours for FED and NFE
• Cost of learning for NFE
• Obstacles to participation in education and training

Researchers /research institutes/students - topics of interest:

• Access to information on learning possibilities and guidance
• Participation in education and training by type, characteristics of the activity (field, distance learning, etc.)
• Reason, use and outcomes of FED and NFE
• Share of job-related or employer-sponsored NFE
• Volume of instruction hours for FED and NFE
• Cost of learning for NFE
• Obstacles to participation in education and training
• Self-reported language skills

International organisations (OECD, UN) - topics of interest:

• Participation in education and training by type, characteristics of the activity (field, distance learning, etc.)

No user satisfaction surveys were carried out for AES 2022.

All variables as requested by the legislation are covered.

Not applicable.

The CAWI design produced data with reasonable accuracy.

• To improve the accuracy of participation of NFE activities, in the questionnaire a maximum of 20 NFE activities was registered (5 courses, 5 workshops, 5 GOJT activities and 5 private lessons). Two of those activities were randomly selected in the questionnaire to answer additional questions. After the data collection, during data processing 5 NFE activities were also chosen by a random procedure.
• To stimulate respondents to participate in the survey, incentives were used. Respondents could win an iPad.
• Where possible, register data are used instead of questions (e.g. birthplace, income).

See item 13.2.1.

For sampling and weighting, see item 18.1.

All calculations are carried out in SPSS25 using Complex samples procedure with a weighting factor of the inverse of the inclusion probability (due to the overrepresentedness of persons in the young age group 18-24 years).

Standard Error = s / √ n where s =standard deviation, n=sample size

Coefficient of Variation = (square root of the estimate of the sampling variance) / (estimated value)

Confidence interval: where p=sample proportion, n=sample size, and z=1.96 (for 95% confidence)

See table 13.2.1 “Sampling errors - indicators for 2022 AES key statistics” in annex “NL - QR tables 2022 AES (excel)”.

See items 13.3 1 to 13.3.5.

We oversampled the youngest age group 18-24 years to meet the precision criteria on the indicator of participation in formal education of this age group.

We had a relatively small percentage of ineligible cases (0.5%). These were individuals that had moved or persons who died between the reference date of the sampling frame and the moment of the survey. There were no individuals that were out-of-scope (individuals who were not in the target population).

In AES 2016 we had a CAWI/CATI survey and we had problems with the providers of telephone numbers. There were relatively many unknown or untraceable sample units (phone number that does not belong to the person we intended to speak). We did not encounter this problem (unknown or untraceable sample units) in the current wave of the AES where we applied CAWI only.

See table 13.3.1.1 “Over-coverage - rate” in annex “NL - QR tables 2022 AES (excel)”.

Not applicable.

The selection procedure applied for selection of the five non-formal education activities (including the two random ones) minimised measurement errors, as the random selection is carried out automatically in the questionnaire itself and in the data processing stage. 

The long reference period can be problematic if the detailed questions concern a randomly selected NFE-activity that took place early in the reference period. As a consequence, e.g. the variables that contain estimates of costs and time spent related to learning activities are probably connected with measurement errors. To help the respondent, estimates are allowed when the exact answer is not known. For the costs, some categories are provided for answering the questions (e.g. less than 100 euros, 100-500 euros, instead of .... euros).

See items 13.3.3.1 and 13.3.3.2.

See table 13.3.3.1 “Unit non-response - rate” in annex “NL - QR tables 2022 AES (excel)”.

See table 13.3.3.2 “Item non-response rate” in annex “NL - QR tables 2022 AES (excel)”.

To ensure the quality of the automatic coding of the educational activities (field) during data processing a classification file of learning activities was used. A manual review was carried out afterwards to minimise coding errors.

Not applicable.

See items 14.1 and 14.2.

See items 14.1.1 and 14.1.2.

The final day of the fieldwork was 31st of March 2023. The first national results are planned to be published in the first quarter of 2024. So the time lag is 9 to 12 months.

The fieldwork ended on 31st of March 2023. We expect to publish our final results on AES 2022 in December 2024 at the latest.

See item 14.2.1.

Delivery of data was in accordance with the target date for delivery. 

Punctuality of publication: Not applicable.

See table 14.2 “Project phases - dates” in annex “NL - QR tables 2022 AES (excel)”.

See table 15.1 “Deviations from 2022 AES concepts and definitions” in annex “NL - QR tables 2022 AES (excel)”.

No additional variables related to COVID-19 were collected.

Not applicable.

The research design of AES 2022 differs from that of AES 2016 and 2011. In 2022 CAWI only mode was used. In 2016 a mixed-mode design was implemented with CAWI and CATI mode. In 2011, only CATI mode was used. Furthermore, the questionnaire became a bit less complex due to a less complex way of asking the questions about education. Before 2016, the whole history of followed educational activities was collected. This is not the case anymore for AES 2016 and AES 2022. We also asked fewer questions about the labour market position of the respondents. Some questions we asked in AES 2016 turned out to be redundant and were deleted in AES 2022.

See table 15.2 “Comparability - over time” in annex “NL - QR tables 2022 AES (excel)”.

Not applicable.

The participation rates in formal and non-formal learning are different for AES compared to the LFS 2022, whereas patterns are similar. The participation rates in formal and non-formal learning activities are higher for AES 2022 than for LFS 2022. The different results are at least partly due to differences in data collection mode(s) and questionnaire design. In addition, the definition of non-formal education in LFS is different (guided on the job training is not taken into account in the LFS).

According to CVTS 6 (2020) the Dutch participation of employees in CVTS courses was around 39% in 2020. Selecting for persons employed in enterprises who participated in courses (similar to CVTS), the participation for AES 2022 is 42%. The difference between AES 2022 and CVTS 2020 results may (at least) be caused by differences in research design, target population (persons vs. enterprises) and reference year (2020 vs. 2022).

The populations in AES and LFS are comparable when crossed by sex, age and education level. In some groups (lowest ISCED levels (0-2) in the age groups 18-24 years and 25-54 years) there are small differences between AES and LFS. This is related to confidence margins on both samples.

See table 15.3 “Coherence - cross-domain” in annex “NL - QR tables 2022 AES (excel)”.

Not applicable.

Not applicable.

AES results for a given data collection round are based on the same microdata and results are calculated using the same estimation methods, therefore the data are internally coherent.

Involved staff for project management, sample design and weighting, questionnaire development, data processing, data analysis and quality report: 2 FTE

Average duration of the CAWI survey: 19.1 minutes (standard deviation: 9.7 minutes)

Not applicable.

Not applicable.

Not applicable.

1) Name and a short description of the sampling frame / register used:

The Municipal basic registration of population data. This registration contains information of all people who are registered at a municipality. The municipalities own the data, but the Ministry of the Interior is responsible for the regulation. Statistics Netherlands is allowed to use it, under strict conditions.

2) Description of the sampling method used and the method used for determining the sample size and sample selection:

The sampling frame from which the AES sample is taken is based on the addresses in the Municipal basic registration of population data. The AES utilised a two-stage sampling procedure. The first stage was a stratified systematic probability proportion to size (PPS) sample of municipalities. This means that municipalities are selected within each stratum with probabilities proportional to their population numbers (municipalities in random order). For the stratification the country is split into 56 strata (areas) that are combinations of 40 Corop areas and 13 interviewing areas (Corop = Committee for the Co-ordination of a Regional Research Programme). Subsequently, sample sizes of persons for the selected municipalities are determined. In the second stage a random sample of persons is drawn from the selected municipalities. The sample size was determined as follows: 

The starting point was a minimal AES response in the two sub populations for which the precision requirements are set in the EU regulation (Annex II): persons in the age of 18 to 24 years and persons in the age of 25 to 69 years. To calculate the minimal sample size the formula f(N)=a√N+b was used. For the indicator formal education 18 to 24 years: 200*square root(1.530.693/1000000)+1500, N is the population in the corresponding age group based on LFS 2021. For the indicator non-formal education 25 to 69 years: 400*square root(10.079.617/1000000) +2000, with N the population in the corresponding age group based on AES 2016. In the calculation of the (maximum) standard error in the two age groups participation rates of 68.9 % (based on LFS 2021) in formal education and 61.5% in non-formal education (based on AES 2016) were used. Based on the calculations above, the minimal net sample sizes were 1,747 in the age group 18 to 24 years and 3,270 in the age group 25 to 69 years. In total 5,017 responses were expected. Based on earlier experience from similar surveys in the Netherlands, it was expected that 28.0% of the persons would respond to the CAWI-survey. The gross sample size was 17,920 persons and 30% of the respondents answered to the questionnaire, i.e. 5,384 persons. The response rate in the age group 25 to 69 years was higher than expected (the response rate was 32.3%) and the response in the age group 18 to 24 years was lower than expected (26.4%).

3) Description of the extrapolation or weighting procedures used to gross up the results in the net sample to the (target) population, discussing the different steps taken or factors applied to the design weighting to take into account the (post-)stratification, adjusting for unit non-response, etc.:

To build the weighting model, a number of potential weighting variables were considered. These potential weighting variables were chosen in such a way that they were expected to be related to skewness in the response and/or related to a number of target variables, e.g. participation in (non-)formal education, highest attained education level. The final weighting model corrects as much as possible for skewness of all potential weighting variables in the response (including those not used). Adding more weighting variables no longer has a significant effect on the estimates of the target variables. Since the inclusion probability is not equal for every element in the sample (the 18-24 group is overrepresented in the sample), the starting weight is calculated by taking the inverse of the inclusion probability.

Using a linear weighing method, a correction factor is calculated. This calibration process adjusted the starting weights so that the estimated totals in the migration background group (10), the gender-age group (2x11), the socio-economic category (7), degree of urbanisation (5), place in household (8) and the welfare quintile (5), fitted the corresponding population values. The final AES weight is equal to the starting weight times the correction factor.

See table 18.1 “Source data” in annex “NL - QR tables 2022 AES (excel)”.

Every 6 years.

See also table 18.1 “Source data” in annex “NL - QR tables 2022 AES (excel)”.

Questionnaire added as annexe "NL - AES 2022 - Survey questionnaire"

• In order to pre-empt mistyping errors there are checking functions integrated in the data input program Blaise itself, which check certain relations between data elements. Thus, correct data entry, coding and routing is forced by Blaise for most of the variables and relations between variables.
• The Dutch AES questionnaire differs from the EU questionnaire in the design of questions, variables and categories of variables. The data collected from the Dutch AES questionnaire therefore has been coded into the variables in the EU questionnaire. The process for creating the EU variables has been checked in order to ensure that this process does not create errors. Checking of valid values for the main EU variables has been done. Also checking that categories of variables in the Dutch questionnaire were correctly coded into categories in the EU questionnaire was done.
• The validation rules from the AES manual were also built into our data processing process so that we could understand and resolve any errors in a timely manner.
• All changes done to the micro data is documented at Statistics Netherlands. The documentation is only available in Dutch language. Original data is also saved at Statistics Netherlands.
• A method for estimating final weights was developed and implemented (see under item 18.1 Source data).
• Because of the CAWI mode with checking functions included, there were few errors in the post-data collection process.

The sampled respondents were invited by letter to fill in the questionnaire on the internet (CAWI). Statistics Netherlands received 30% response via internet. The CAWI questionnaire included some simple checks (e.g. male+female=total). The data were automatically checked according to the available checking rules in the manual, which were programmed in BLAISE (Manipula). The inevitable occurring errors were corrected by editing of data, mainly using general correction rules. In some cases manual editing was needed.

Not applicable. 

Not applicable. See table 18.5.1 “Imputation - rate” in annex “NL - QR tables 2022 AES (excel)”.

Not applicable.

Not applicable.

None.